pm-ship
Skill2 starsChanged 40 days agoMITView on GitHub
Phase 6, Ship. Run the pre-ship checks for real, record test runs and the deployment, and promote delivered stories to verified. Use when the user says "/pm-ship", "deploy this", "release", "is this ready to ship", or after a milestone's build gate is approved.
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 11
Every item here is executed and its result recorded.
- Very specific: line 39
Run the security-review skill against the diff, plus whatever the stack provides (dependency audit, secret scan).
- Very specific: line 51
Every state in every screen's states_json is actually implemented, not just the happy path.
- Very specific: line 57
No raw color, size or radius in the code repo that should have been a token from standards/tokens.css.
- Very specific: line 85
If nothing tests it, it stays done.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
/pm-ship: checks that run, not a checklist that gets ticked
Ship comes after /pm-edit. Every delivered epic needs its edit gate approved first, and
./pm check fails until it is. Tests prove it works. The edit pass proves it is good.
Every item here is executed and its result recorded. A checkbox nobody ran is worse than no checkbox.
1. Correctness
./pm check # invariants: orphans, stale screens, open gatesThen each epic's Prove it end to end steps, exactly as written in its doc. A step that no longer matches the product is a doc bug, fix the doc in the same pass.
Then the project's own test commands. Record each run:
./pm add test_run kind=unit tool=<runner> passed=<n> failed=<n> coverage_pct=<n> ref=<sha> \
covers=STORY001,STORY004
./pm add test_run kind=e2e tool=<runner> passed=<n> failed=<n> ref=<sha> covers=STORY002covers is the load bearing field. It lists the story codes this run actually exercises,
and it is what lets a story reach verified.
Failing tests block the ship. If the user wants to ship anyway, that is a gate skip with a reason on the record, not a quiet omission.
2. Security
Run the security-review skill against the diff, plus whatever the stack provides
(dependency audit, secret scan). Record it:
./pm add test_run kind=security tool=<tool> passed=<n> failed=<n> ref=<sha>Anything unresolved becomes a task before ship, or an explicitly accepted risk written into the area TRD's deferred section. Never a silent pass.
3. Completeness
- Every state in every screen's
states_jsonis actually implemented, not just the happy path. Query them, do not trust memory. - Responsive down to 375px.
- Keyboard reachable, visible focus,
prefers-reduced-motionrespected. - Both themes render correctly, including the unstamped system-default state.
- Anti-slop check from
CLAUDE.md. - No raw color, size or radius in the code repo that should have been a token from
standards/tokens.css. Grep for hex values outside it.
./pm q "SELECT code,name,states_json,design_status FROM screen WHERE design_status!='approved'"
./pm q "SELECT code,title FROM work_item WHERE level=1 AND status='done'"Any screen not approved blocks the ship or gets explicitly cut from this release.
4. Deploy and record
./pm add deployment env=production version=<tag> sha=<sha> url=<url> status=liveSmoke test the live URL. If it fails, record the rollback as another deployment row with
status=rolled_back rather than deleting anything. The history is the value.
5. Promote stories
A story reaches verified only when a passing test run names it, and ./pm check
enforces that rather than trusting it:
./pm set STORY001 status=verified # needs a passing test_run whose covers lists STORY001If nothing tests it, it stays done. That gap is honest and useful, and the Quality view
lists every story sitting in it. Relabelling will just fail the next ./pm check.
6. Close the gate
./pm gate open ship <version>
./pm gate approve ship --subject <version> "checks green, deployed and smoke tested"
./pm phase nextAsk before anything outward facing
Deploys, DNS changes, public posts, anything users will see. Confirm with a form first, even when the user asked for it earlier in the session. Approval in one context does not carry to the next one.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

